CALIBRATING PHYLOGENIES ASSUMING BIFURCATION VERSUS BUDDING ALTERS INFERRED MACROEVOLUTIONARY DYNAMICS OF BIVALVE FAMILIES

Assumptions about the nature of lineage diversification can profoundly affect the calibration of phylogenetic nodes, and thus the quantification of evolutionary dynamics. Molecular phylogenies assume a purely bifurcating process, where nodes are calibrated using the older of the two daughter lineages. However, in many high taxa, lineages are inferred to diversify via a budding process, in which a parent lineage persists following the origin of a daughter lineage, and nodes are calibrated using the age of the younger lineage. The effects of these two assumptions on dating phylogenies are being increasingly evaluated, but changes to inferred macroevolutionary dynamics remain poorly understood. Here, we use the extensive fossil record of bivalve molluscs to test how assumptions about the process of evolution affect resulting macroevolutionary patterns. We time-calibrated 91% of nodes in a phylogeny of 97 extant bivalve families using 86 calibration points, which range in age from 2.59 to 485 Ma. The resulting phylogenies show that budding produced calibrations greatly diminishes the summed duration of inferred “ghost lineages,” from 6.76 billion yrs (Gyr; bifurcation model) to 1.00 Gyr (budding model). Adding 30 extinct paraphyletic stem families from the literature, deep splits are pushed back, and ghost lineages total 7.83 Gyr (bifurcation) and 1.92 Gyr (budding). Lineage-through-time plots from phylogenetic data scaled under a bifurcating model of evolution pushes more of bivalve diversification into the Paleozoic, and steepens and shortens the group’s response to mass extinction events, relative to a budding model. Overall, our results show that prior consideration of the hypothesized diversification process within a given clade is essential when node-calibrating phylogenies, and that for a major clade with a robust fossil record, at least, budding appears to be the most appropriate evolutionary model.